Functional Haplotypes in the ADIPOQ Gene are Associated with Underweight, Immunosuppression and Viral Suppression in Kenyan HIV-1 Infected Antiretroviral Treatment Naive and Experienced Injection Substance Users

Abstract Background Human immunodeficiency virus and injection substance use have an influence on genes and gene expression. These effects could be beneficial or detrimental in defining disease outcomes. Adiponectin gene is key in modulating metabolic and immunoregulatory functions. Understanding the effects of human immunodeficiency virus and injection substance use on the gene in the context of antiretroviral therapy is important for predicting disease outcomes. Methods This cross-sectional genetic study determined polymorphisms in the promoter region of adiponectin gene. Two variants were analyzed: rs2241766 and rs266729. Polymorphisms were associated with clinical markers of disease outcome; underweight, immunosuppression and viral suppression. The variants were genotyped via random fragment length polymorphism. Result GC haplotype was associated with higher odds of having underweight (OR, 2.21; 95% CI, 1.83–4.60; P=0.008 vs. OR, 2.30; 95% CI, 1.89–4.71; P=0.006) in antiretroviral treatment - naive and experienced injection substance users and immunosuppression (OR, 1.90; 95% CI 1.67–3.98, P=0.041) in naive. Bonferroni correction revealed GC haplotype carriers only to have low body mass index in both naive (median, 14.8; IQR, 3.2 kg/m2; P=0.002) and experienced (median, 15.2; IQR, 3.2 kg/m2; P=0.002) injection substance users. Circulating total adiponectin levels were higher in naive (median, 19.5; IQR, 7.9 µg/ml) than - experienced (median, 12.0; IQR, 4.4 µg/ml) injection substance users (P=0.0001). GC carriers presented with low serum adiponectin levels in both study groups. Conclusion The study revealed haplotypes of adiponectin gene at loci rs2241766 and rs266729 that could determine disease outcomes in human immunodeficiency virus -1 antiretroviral treatment- naive and experienced injection substance users.


INTRODUCTION
Of an estimated 13 million people who inject illicit drugs in the world, 1.7 million are HIV infected (1). In Africa, ~630,000 people are injection substance users (ISUs) of whom 13.6% are living with HIV (2). In Kenya, over 50,000 ISUs have been identified, and the number is on the increase predominantly in coastal, urban, peri-urban areas and cities (3). The prevalence of HIV infection in Kenyan ISUs is ~18.3% with higher incidence of 18.7% mainly at the coast linked to high rates of injection substance use (4).
Adiponectin gene (ADIPOQ) encodes adiponectin (5), a key adipokine regulating carbohydrate and fat metabolism, immune and inflammatory responses (6). ADIPOQ gene is ~15.8 kb spanning across three exons on chromosome 3q27 (7). ADIPOQ gene is highly expressed in adipocytes, and moderately in renal epithelial cells, macrophages and CD4+ T cells (8,9). Genetic variations in the ADIPOQ gene promoter loci -676 to +41 notably rs2241766, rs266729, rs17300539, rs16861210, rs6444175 and rs1501299 are the most extensively studied in relation to disease outcomes across populations (10). The rs2241766 variants are associated with metabolic syndrome, obesity and hypertension (10,11), while rs2241766 and rs266729 genotypes and haplotypes correlate with BMI in diabetics (12). Moreover, rs266729 is associated with nonalcoholic fatty liver disease in smokers (13), and onset of type-2 diabetes mellitus (14). In HIV infection, rs266729 GG variant is associated with low CD4+ T cell counts among patients on antiretroviral treatment (ART) (15). Also, a review of the two loci using ENCODE, haploreg, HapMap and regulome databases revealed rs2241766 to be a synonymous SNP with enhancer histone markers in blood while rs266729 as an intronic variant associated with ADIPOQ gene expression.
It appears that ADIPOQ variants mediate disease outcomes by controlling gene expression. For instance, in vitro studies in recombinant adiponectin treated HEK293 cells deficient in metabolic activity showed increased expression of the ADIPOQ gene in skeletal muscles (16). Moreover, several in vivo studies also showed that polymorphisms in the ADIPOQ gene influence gene expression which translates into altered circulating adiponectin levels (17). In addition, ADIPOQ TG haplotype at rs2241766 and rs266729 loci, and at +45 and +276 loci were associated with low circulating adiponectin levels in insulin resistant Korean and Indian patients, respectively (18).
Altogether, these findings suggest that haplotypes are better predictors of circulating adiponectin levels and disease progression. Additionally, underweight, immunosuppression and viral suppression are key clinical markers of disease progression in both HIV-infected and noninfected ISUs (19). However, there is limited data on how ADIPOQ gene polymorphisms influence these markers of disease progression among HIV-1 ISUs. We have previously shown that serum adiponectin levels are low in HIV-1 ART-naive ISUs from coastal Kenya (20). In addition, injection substance use leads to epigenetic modifications in gene expression and causes alterations in the DNA sequences (17). Furthermore, HIV infection promotes a repertoire of up-regulations and/or down-regulations of human gene expression (21). Consequently, HIV-1 ISUs provide a unique population in which gene regulation can be studied. Therefore, we examined associations of ADIPOQ gene variants at rs2241766 and rs266729 loci with underweight, immunosuppression and viral suppression among HIV-1 ART naive and experienced Kenyan ISUs.

Study site and population:
This cross-sectional study was conducted among HIV-1 ART-naive and -experienced ISUs from Mombasa, a coastal city in Kenya. Detailed description of the study site and population, clinical and adiponectin measurements are presented in our recent publications (22). In the current study, the ISUs were stratified into ARTnaive and -experienced individuals. ART-naive ISUs were individuals testing positive for HIV-1, reporting having never used ART and having no AIDS defining illnesses. ART-experienced were individuals testing positive for HIV-1, currently on ART and having no AIDS defining illnesses. The clinical measures of underweight, immunosuppression and viral suppression were used in examining ADIPOQ gene regulation of HIV-1 infection in ISUs. Underweight was defined by BMI <18.5 kg/m 2 (23). Although WHO classifies HIV infection into four immunologic stages based on CD4+ T cell count (24), immunosuppression was based on CD4+ T cell count <500 cells/µl. Viral suppression was also defined as per the WHO using >1000 HIV RNA copies/ml (25).
CD4 T cell enumeration: CD4+ T cell counts were determined in an automated fashion using the BD FACSCalibur flow cytometer (Becton-Dickinson™, Franklin Lakes, USA). Briefly, 5.0 µl of EDTA blood samples were placed in a tube and RBC lysis buffer added. After 5 minutes incubation, the cells were washed and fluorescenttagged antibodies (anti-CD3, anti-CD4, and anti-CD45) were added. The cells were incubated for 30 minutes after which the samples were washed and the CD4+ T cells enumerated on the flow cytometer. HIV-1 viral load: HIV-1 viral loads were determined using the automated Abbott m2000 System according to the manufacturer's instructions (Abbott Molecular Inc., Illinois, USA). Briefly, RNA was extracted from 0.2 ml serum samples and reverse-transcribed into cDNA. The cDNA was amplified using HIV-1-specific and internal control primers. Fluorescence intensity of the HIV-1 probe was converted into viral loads by the analyzer. Serum adiponectin measurement: Total adiponectin concentrations were determined in serum using commercial ELISA reagents according to the manufacturer's protocols (R&D Systems, Minneapolis, MN, USA). SNP selection strategy: SNPs rs2241766 and rs266729 were selected based on >5% distribution of alleles in HapMap populations (26,27). Also, location/co-localization with putative transcription factor binding sites and reported association with adiponectin levels was used. Furthermore, other adiponectin related traits and their closeness in genomic proximity were considered. DNA extraction and genotyping: DNA was extracted from dry blood spot (Whatman plc) samples using QiaAmp™ DNA Mini Kit (Qiagen Inc., Valencia, USA) according to the manufacturer's protocols. Genotyping was carried out by PCR-restriction fragment length polymorphism (RFLP). PCR of DNA segments, containing rs2241766 and rs266729, was performed using the following primers: rs2241766, forward primer 5'-GCAGCTCCTAGAAGTAGACTCTGCTG -3' and reverse primer 5'-GCAGGTCTG TGATGAAAGAGGCC -3'; rs266729, forward primer 5'-GGTGGACTTGACTTTACTGG -3', and reverse primer 5'-TAGAAGCAGCCTGGAGAA -3'. PCR was carried out in a final volume of 50 µL containing: 5 µL 10X PCR buffer (KEMTAQ ® ), 4 µL MgCl2, 2 µ dNTP, 10 pmol of each primer, 0.5 µL Taq DNA polymerase, 32.5 µL ddH2O, and 4 µL genomic DNA. The cycling conditions for rs2241766 and rs266729 were 94°C for 2 minutes, 35 cycles of 94°C for 15 seconds, 60°C for 30 seconds, and 72°C for 30 seconds. This was followed by a final extension at 72°C for 10 minutes and a pause at 4°C. The PCR products were separated on a 2% UltraPure™ agarose gel and visualized under UV light. PCR products were digested with 10 U/µL restriction enzymes (New England Biolabs, Ipswich, USA). SNPs rs2241766 (SmaI), and rs266729 (HhaI) were genotyped by PCR-RFLP. The digested products were separated on a 2% NuSieve® 3:1 agarose gel (NuSieve® GTG® Agarose). The digests on agarose gel were read and scored separately by two different trained technologists-in the case of disagreement a third trained research assistant was called in. Samples that did not amplify for both variants were removed during analysis. Data analysis: Statistical analysis was performed using Statistical Package for the Social Sciences (SPSS) version 24.0 (SPSS Inc, Chicago, USA) and GraphPad Prism version 6.0 (GraphPad, California, USA). Continuous data such as age, height, weight, BMI, CD4+ T cell counts, log10 HIV-1 RNA copies and adiponectin levels were compared between ART-naive and -experienced ISUs using the Mann-Whitney U test. Discrete data and categorical data were compared between the study groups using the Fisher's exact tests. Hardy-Weinberg equilibrium was determined using the Online Encyclopedia for Genetic Epidemiology studies software at www.oege.org/software/hwe-mr-calc.shtml.
Haplotype distribution was determined using the HPlus haplotype analysis software-http://cdsweb01.fhcrc.org/HPlus/ and Haploview-www.broadinstitute.org/haploview. Kruskal-Wallis test was used to determine acrosshaplotype differences, and where significant, posthoc Mann-Whitney U tests with the Bonferroni correction were performed with cut-off set at P<0.002. Binary logistic regression analyses were used to determine the associations of haplotypes with underweight, immunosuppression and viral suppression. In these regression models, we controlled for age, gender, duration and frequency of drug injection in the ART-naive; and age, gender, duration and frequency of drug injection, and duration of ART use in the ART-experienced ISUs. All tests were two-tailed and P<0.05 was used for statistical inferences. Ethics approval and consent to participate: Ethical approval for this study was obtained from Kenyatta University Ethics Review Committee (Protocol: PKU019/116/2012) and Masinde Muliro University of Science and Technology Institutional Ethical Review Committee (Protocol: MMU/COR-403012-V27).
Written informed consent was obtained from each participant before enrolment. All HIV-1 infected ART-naive study participants were referred for further treatment and care at the HIV Comprehensive Care Clinics

RESULTS
Demographic and clinical profiles of the study participants: The demographic and clinical characteristics of the study participants are summarized in Table 1  Haplotype frequencies: Haplotype frequencies are shown in Figure 1A.

Association of haplotypes with underweight:
Previous studies showing that underweight is linked to low serum adiponectin levels and predicts wasting in HIV infection and substance use (28), led us to examine the relationship of ADIPOQ haplotypes and underweight. Associations of ADIPOQ haplotypes with underweight are shown in Figure 1B and C. The GC haplotype was associated with higher probability of having underweight in both ART-naive (OR, 2.21; 95% CI, 1.83-4.60; P=0.008) and -experienced (OR, 2.30; 95% CI, 1.89-4.71; P=0.006) ISUs.

Association of haplotypes and immune suppression:
Since adiponectin promotes CD4+ T cell depletion and decreased CD4+ T cell numbers predicts progression of HIV infection (29), we examined the association between ADIPOQ haplotypes and immunosuppression. Associations of ADIPOQ haplotypes with immune suppression are shown in Figure 1D and E. This analysis revealed that GC (OR, 1.90; 95% CI 1.67-3.98, P=0.041) and GG (OR, 2.02; 95% CI 1.61-4.12, P=0.017) haplotypes were associated with a higher probability of having immunosuppression in ARTnaive and -experienced ISUs, respectively.
Association of haplotypes with viral suppression: The viral load is associated with serum adiponectin and progression of HIV infection (30); hence, we examined whether ADIPOQ haplotypes were associated with viral suppression. Associations of ADIPOQ haplotypes with viral suppression are shown in Figure 1F and G. Analysis of the haplotypes revealed that GG (OR, 2.11; 95% CI 1.49-4.11, P=0.011) and TG (OR, 1.92; 95% CI 1.51-3.97, P=0.016) haplotypes were associated with a higher probability of having viral suppression in ART-experienced ISUs only.

DISCUSSION
This study focused on two variants of the ADIPOQ gene: rs2241766 and rs266729. Both variants occur in regions where transcription factors bind to the gene (11). We compared haplotype distribution between the two study groups. Four possible haplotypes were generated-GC, GG, TC, and TG. Rate of GC carriage was lower in the -naive compared to -experienced ISUs. In contrast, TG carriers were more in HIV-1 ART-naive ISUs compared with -experienced suggesting that ADIPOQ haplotypes could be influencing disease outcomes in HIV-1 ISUs. Previous studies have reported ADIPOQ haplotypes modulating obesity and type 2 diabetes (31). There is limited data on ADIPOQ haplotype modulation of disease outcomes in infectious diseases including HIV-1. This study is the first one to describe ADIPOQ haplotypes in HIV-1 infected ISUs.
Consistent with our previous study (20), ART-naive ISUs had higher serum adiponectin levels suggesting that ART could directly or indirectly be causing lower levels of circulating adiponectin. Previous studies have associated ART use with low levels of serum adiponectin in HIV infected individuals. For instance, stavudine has been shown to lower serum adiponectin up to 3.8 folds (32). In addition, a decrease in circulating adiponectin concentration has been reported in individuals on HAART (33). NRTIs have also been found to modulate lipodystrophy in HIV patients which is associated with altered circulating adiponectin levels (34). Evidence from genome wide association studies has revealed derangements that arise in genes involved in adipocyte differentiation and lipid metabolism on exposure to ARVs (35). Contrastingly, Morimoto et al., showed that hypoadiponectinemia exhibited by HIV-1 infected patients is not influenced by ART use (36). Taken together, these studies suggest a direct or an indirect effect of ART on the circulating levels of adiponectin.
Consistent with lower BMI, the GC haplotype was associated with higher odds of having underweight in both ART-naive and -experienced ISUs. The GC haplotype carriers also had a higher chance of having immunosuppression in both the ART-naive and -experienced individuals. These findings suggest that ADIPOQ haplotypes modulate HIV-1 infection and treatment outcomes. Indeed, ADIPOQ haplotypes have been previously associated with disease progression in several pathological conditions. For instance, the TG haplotype has been linked with disease outcomes for patients with insulin resistance, coronary heart disease and obesity (18). However, very few studies have investigated ADIPOQ haplotypes in the context of infectious diseases. Kato et al reported ADIPOQ haplotypes to be modulating serum adiponectin levels in HIV infected individuals (15), but no study has yet described ADIPOQ haplotypes in the context of HIV-1 ISUs. Our findings therefore imply that the GC haplotype could be driving deleterious disease outcomes in HIV-1 infected ISUs. Nonetheless, the association of ADIPOQ haplotypes with BMI in the current study should be interpreted with caution as age and gender of an individual correlate with BMI such that younger individuals and females are more likely to have a lower BMI (37).
Overall, our findings appear to predict that HIV-1, injection substance use, antiretroviral treatment and polymorphisms in the promoter of ADIPOQ gene modulate the circulating adiponectin levels. Furthermore, variations in the promoter region of the ADIPOQ gene may affect the affinity, avidity and magnitude for which the gene binds with its transcription factors both cis and trans upstream so as to initiate transcription. Therefore, this modulates transcription and subsequently translation altering the circulating adiponectin levels. This in turn affects disease outcomes in HIV-1 infected ART-naïve andexperienced ISUs.
In conclusion, this study revealed associations between ADIPOQ gene variant haplotypes and clinical markers in HIV-1 ISUs. ADIPOQ polymorphisms can thus be used as surrogate markers in HIV-1 infected ISUs. and staff of Bomu Hospital for their support during the study.